in our project we uses jBPM in the cluster environment with two servers. When the process is executed on the server1 and the node fails, the process stay in the last commited node and must be signalled to continue manually.
Is there exists some solution to prevent this situation? Is there any possibility to automatically recognize these "dead" processes to notify the operator that can signal the process?
recognizing 'dead' processes by the engine is very difficult. How can it differentiate from a process that 'normally' is in a specific state...
Normally, if it fails, the 'signalling' application/service/user/... gets an exception and will have to retry. That is the only place where it is really known how to respond.